Processed Data for Graph-Based Trajectory Modeling of Type 2 Diabetes

This dataset contains processed and derived features used in the study "Graph-Based Trajectory Modeling for Progression-Aware Type 2 Diabetes Subtyping". The data were derived from the MIMIC-IV v2.2 database and include:- Longitudinal laboratory measurements- Medication exposure indicators- Static patient features- Learned temporal embeddings- Cluster assignments used in analysis Due to MIMIC-IV data use restrictions, this dataset contains only processed, de-identified, and non-reversible features. Raw MIMIC-IV data are available via PhysioNet under a data use agreement. This dataset is intended solely for research and reproducibility purposes.本数据集包含用于研究《用于进展感知型2型糖尿病分型的基于图的轨迹建模》（"Graph-Based Trajectory Modeling for Progression-Aware Type 2 Diabetes Subtyping"）的经处理与衍生得到的特征。 该数据源自MIMIC-IV v2.2数据库，涵盖以下内容： - 纵向实验室检测指标 - 药物暴露指示特征 - 静态患者特征 - 学习得到的时序嵌入（Temporal Embedding） - 分析中使用的聚类分配（Cluster Assignment） 由于MIMIC-IV存在数据使用限制，本数据集仅包含经处理的去标识化且不可逆特征。原始MIMIC-IV数据可通过PhysioNet平台签署数据使用协议后获取。 本数据集仅用于科研及研究可重复性验证目的。